Handling Incidents
The incident layer. When something breaks the right people hear about it on a clock, the fix gets dispatched and confirmed, the blameless post-mortem lands within days, its action items get tracked to closure, and alerts that fire without a human action get fixed or deleted.
Your Righthand runs incidents end to end. During one, it posts the four-line update on a cadence (what is broken, who is affected, what is being done, when the next update comes), diagnoses the failure, dispatches the revert or fix to a coding agent, reviews the diff, and confirms the fix in production before it says "resolved". Afterwards it writes the blameless post-mortem within days, chases every action item to closure, and every Tuesday prunes the alerts nobody acted on.
What's in the pack.
3 skills, installed together — your Righthand arrives already equipped for the job.
- Operations
Incident Communications
Use when something is broken in production or during incident follow-through: severity levels, the four-line update posted on a clock, the shape-not-stack-trace rule for the founder, driving a revert or fix through a coding agent to confirmed, the blameless post-mortem within days, action items tracked to closure, and alert hygiene. Maintains Knowledge/Organizations/incident-protocol.md and Knowledge/Organizations/incidents/.
- Communications
Sending Communications
Use whenever you need to send an outbound SMS, email, Slack message, or voice call from your own number/address. Your final response text is internal logging only — the ONLY way to talk to the outside world is the agent CLI.
- Engineering
Using Coding Agents
Delegate bounded engineering implementation to an asynchronous Codex or Claude coding job, answer only judgment questions, and independently review the resulting diff against user acceptance scenarios.
Hire a Righthand
ready for this job.
Pick Handling Incidents at signup and your Righthand shows up with every skill in the pack already installed.